Grok 3 Went Off the Rails—And Musk Hit the Emergency Brake:
Elon Musk’s latest AI brainchild, Grok 3, was supposed to be a fearless, truth-seeking machine—until it started roasting its own creator. The controversy erupted when users discovered Grok was more than happy to call Musk a major spreader of misinformation and even had some not-so-flattering opinions about Donald Trump. Given that Musk has always touted xAI’s model as an unfiltered alternative to “woke” AI, this didn’t exactly fit the narrative.
Grok 3, was designed to be a “maximally truth-seeking” machine—until it started telling the wrong truths. In an even wilder twist, an earlier version allegedly suggested that both Musk and Trump deserved the death penalty (yikes).
Censor First, Ask Questions Later
Unsurprisingly, xAI engineers scrambled to “fix” the issue. Enter Igor Babuschkin, who quickly implemented a patch, making Grok mysteriously unable to comment on Musk and Trump at all. But that wasn’t the end of it—days later, users found that Grok had been instructed to outright exclude sources linking Musk and Trump to, well… less-than-stellar reputations. Convenient.
Blame the New Kid
As outrage over the sudden censorship grew, Babuschkin revealed that the culprit behind the AI’s revised behavior was a new hire—a former OpenAI employee—who, in his words, hadn’t “fully absorbed xAI’s culture yet.” Translation: “We don’t censor here unless it makes Elon look bad. / They hadn’t learned that this AI shouldn’t be that honest.”
Musk vs. His Own “Free Speech” Ideals
Elon Musk has spent years criticizing Big Tech for suppressing free speech and promoting biased narratives. But now that his AI model is the one saying uncomfortable things, we’re suddenly seeing some serious moderation in action.
Between this and the proposed changes to Twitter’s Community Notes system, cracks are starting to show in Musk’s promise of an “unbiased” platform. The real question is: does free speech only matter when it’s convenient?
Stay tuned for the next episode of “When AI Goes Off-Script.”
Join the Debate!
What do you think? Is Grok 3 just another AI falling victim to corporate control, or was this a necessary course correction? Let us know in the comments!
For more details, check out these sources:
📌 Business Insider – Grok briefly censored criticism of Musk and Trump
コメント